I haven’t implement that feature yet. Can you elaborate how you would set those conditions and to which destinations you’d want to send alerts to?
rodloboz
frankdugan3
I have found LLMs to actually be relatively decent at quickly slopping up complex queries, provided it is aware of the schema. Would be great to have a button that copied a table’s create statement to clipboard (or some similar schema definition) that could be easily pasted into a chat to give it better context.
Or even more interesting, have an opt-in feature where an AI API key could be added to the configuration (or an oauth flow - whatever makes sense), and have an integrated chat for the query builder that only had access to the query, schema, and query errors for debugging.
1
ghannam80
I have some legacy applications that we would like to monitor some criteria in it like if a certain number of transactions rejected in a sliding time window compare it with some threshold and based on it send some defined alert to telegram and stream alerting UI.
Another one for silence detection which is if no entries or low entries within a sliding windows again alert about service down.
Finally , monitor for certain rejection reasons and if number threshold met send an alert.
2
rodloboz
Having a BYOK (bring your own key) AI feature for building queries in natural language is something that’s in the roadmap 
2
rodloboz
Lotus v0.12 / LotusWeb v.11 adds AI-powered query generation 
Generate SQL from natural language, right inside LotusWeb.
How it works:
- Type what you want: “Show customers with overdue invoices”
- AI explores your schema and generates the SQL
- Review, refine, and run
Key details:
- BYOK (Bring Your Own Key) - works with OpenAI, Anthropic, or Google Gemini
- Schema-aware - the AI sees your tables, columns, and even checks actual enum values
- Secure - respects Lotus visibility rules (AI only sees what you see)
- Keyboard shortcut: Cmd/Ctrl+K to open
The AI Assistant is experimental and optional. Perfect for:
- Exploring unfamiliar schemas
- Writing complex JOINs faster
- Getting unstuck on tricky queries
4
frankdugan3
Fantastic start! I gave it a try on a complex query, which resulted in a query error. It doesn’t seem to be conversational, so I’m unclear on the best way to interact. Some ideas:
- It could automatically detect error results and feed that back into the context and try again, sort of agent style.
- Every “generate” could be a new entry in a conversation that stays in the context, along with the previously generated query. (Could have a “Clear”/“Reset” button to start an entirely new query.)
2
rodloboz
frankdugan3
It had all the context it needed, it was just a query that dealt with JSON aggregate type stuff. I’ve always had to go back and forth when working with an LLM in a chat to get it right on previous occasions, even when it has all the schema info. I don’t think it’s reasonable to expect it to succeed with a one-shot for this sort of thing, so I’m looking more for the ability to have some kind of feedback loop for refining the result.
2
rodloboz
